@telepilotco/tdl

This fork of bannerets/tdl does not change source code, but modifies Github Actions build pipeline to extend environments that are supported.

Following environments are supported:

• [x] linux-x64-glibc
• [x] linux-x64-musl
• [x] linux-arm64-glibc
• [x] linux-arm64-musl
• [ ] win-x64 - disabled due to https://github.com/telepilotco/tdl/issues/4
• [x] darwin-x64
• [x] darwin-arm64

tdl

tdl is a fairly simple JavaScript wrapper for TDLib (Telegram Database library), a library to create Telegram clients or bots.

TDLib version 1.5.0 or newer is required.

• Requirements
• Installation
• Getting started
• API
• Types
• Other JavaScript runtimes
• Possible errors

Requirements

• Node.js v12.11.0 or newer (>= v16 recommended)
• The tdjson shared library (libtdjson.so on Linux, libtdjson.dylib on macOS, tdjson.dll on Windows)
• In some cases, a C++ compiler and Python installed to build the node addon[^1]

[^1]: tdl is packaged with pre-built addons for Windows (x86_64), GNU/Linux (x86_64, glibc >= 2.17), and macOS (x86_64, aarch64). If a pre-built binary is not available for your system, then the node addon will be built using node-gyp, requiring Python and a C++ toolchain (C++14 is required) to be installed (on Windows, MSVS or Build Tools). Pass --build-from-source to never use the pre-built binaries. Note that macOS aarch64 binaries aren't tested.

Installation

1. Build TDLib (https://github.com/tdlib/td#building) or install pre-built libraries
2. Run npm install tdl
3. (optional) If you use TypeScript, types for TDLib are installed separately, see the Types section

To use tdl, you need to get TDLib first. The tdjson shared library should be present in the system search paths (otherwise the path to libtdjson can be specified manually).

Note: When building TDLib, the libraries can be installed into the system using cmake --install . (optionally specify the --prefix option, the default is /usr/local) after TDLib has been built successfully. This command may require sudo.

prebuilt-tdlib

Instead of building TDLib from source, you can possibly install pre-built TDLib libraries distributed through the prebuilt-tdlib npm package. To install prebuilt-tdlib for e.g. TDLib v1.8.19, run npm install [email protected] (the available versions of prebuilt-tdlib can be found by running npm info prebuilt-tdlib dist-tags). An example of using libraries from prebuilt-tdlib is present in the section below. The supported systems are x86_64 GNU/Linux, x86_64 & arm64 macOS, and x86_64 Windows. See the README of prebuilt-tdlib for additional information.

Getting started

const tdl = require('tdl')

// If libtdjson is not present in the system search paths, the path to the
// libtdjson shared library can be set manually, e.g.:
//   tdl.configure({ tdjson: '/usr/local/lib/libtdjson.dylib' })
// The library directory can be set separate from the library name,
// example to search for libtdjson in the directory of the current script:
//   tdl.configure({ libdir: __dirname })

// Instead of building TDLib yourself, the aforementioned prebuilt-tdlib can be used as follows:
//   const { getTdjson } = require('prebuilt-tdlib')
//   tdl.configure({ tdjson: getTdjson() })

const client = tdl.createClient({
  apiId: 2222, // Your api_id
  apiHash: '0123456789abcdef0123456789abcdef' // Your api_hash
})
// Passing apiId and apiHash is mandatory, these values can be obtained at https://my.telegram.org/

client.on('error', console.error)

// Aside of receiving responses to your requests, the server can push to you
// events called "updates" which ar received as follows:
client.on('update', update => {
  console.log('Got update:', update)
})

async function main () {
  // Log in to a Telegram account. By default, with no arguments, this function will
  // ask for phone number etc. in the console. Instead of logging in as a user,
  // it's also possible to log in as a bot using `client.loginAsBot('<TOKEN>')`.
  await client.login()

  // Invoke a TDLib method. The information regarding TDLib method list and
  // documentation is below this code block.
  const me = await client.invoke({ _: 'getMe' })
  console.log('My user:', me)

  // Invoke some other TDLib method.
  const chats = await client.invoke({
    _: 'getChats',
    chat_list: { _: 'chatListMain' },
    limit: 10
  })
  console.log('A part of my chat list:', chats)

  // Close the instance so that TDLib exits gracefully and the JS runtime can finish the process.
  await client.close()
}

main().catch(console.error)

The API list of TDLib methods, which are called using client.invoke, can be found at, e.g.:

• https://core.telegram.org/tdlib/docs/annotated.html (possibly outdated)
• or in the td_api.tl file in the TDLib repository.

In the TDLib documentation, the bytes type means a base64-encoded string. int64 accepts either a number or a string, pass string for large numbers. int32, int53, and double are the number JS type. If TypeScript types are installed, note that the types are annotated with jsdoc comments, and the documentation can be browsed directly in the .d.ts file or in the autocompletion menu.

See also https://core.telegram.org/tdlib/getting-started for some basic information on how to use TDLib (tdl handles the authorization part with client.login). Note that the TDLib JSON interface actually sends a @type field, but tdl renames it to _.

Some short examples are available in the examples/ directory.

API

Note: A more exhaustive documentation is available in the TypeScript typings file.

tdl.configure(options: TDLibConfiguration) => void

Configure several parameters such as libtdjson name or verbosity level. This function should be called before tdl.createClient or tdl.execute.

tdl.configure({
  // Path to the library. By default, it is 'tdjson.dll' on Windows,
  // 'libtdjson.dylib' on macOS, or 'libtdjson.so' otherwise.
  tdjson: 'libtdjson.so',
  // Path to the library directory. By default, it is empty string.
  libdir: '/usr/local/lib',
  // Verbosity level of TDLib. By default, it is 2.
  verbosityLevel: 3,
  // Experimental option. Disabled by default.
  useNewTdjsonInterface: false
})

Some examples:

• tdl.configure({ tdjson: '/root/libtdjson.so', verbosityLevel: 5 })
• tdl.configure({ tdjson: 'libtdjson.dylib.1.8.6', libdir: '/usr/local/lib' })
• tdl.configure({ libdir: __dirname })
• tdl.configure({ tdjson: require('prebuilt-tdlib').getTdjson() })

The path concatenation of libdir + tdjson is directly passed to dlopen (Unix) or LoadLibrary (Windows). Check your OS documentation to find out where the shared library will be searched for.

tdl.createClient(options: ClientOptions) => Client

Create a TDLib client.

const client = tdl.createClient({
  apiId: 2222, // Your api_id
  apiHash: '0123456789abcdef0123456789abcdef' // Your api_hash
  // ... other options ...
})

The interface of the options that can be passed here:

type ClientOptions = {
  apiId: number, // Can be obtained at https://my.telegram.org
  apiHash: string, // Can be obtained at https://my.telegram.org
  databaseDirectory: string, // Relative path (defaults to '_td_database')
  filesDirectory: string, // Relative path (defaults to '_td_files')
  databaseEncryptionKey: string, // Optional key for database encryption
  useTestDc: boolean, // Use test telegram server (defaults to false)
  tdlibParameters: Object, // Raw TDLib parameters
  // Advanced options:
  bare: boolean,
  skipOldUpdates: boolean
}

Of these fields, only apiId and apiHash are required. Any other field can be omitted.

The tdlibParameters option is described in https://core.telegram.org/tdlib/docs/classtd_1_1td__api_1_1tdlib_parameters.html.

By default, in tdl, tdlibParameters is set to:

tdlibParameters: {
  use_message_database: true,
  use_secret_chats: false,
  system_language_code: 'en',
  application_version: '1.0',
  device_model: 'Unknown device',
  system_version: 'Unknown',
  enable_storage_optimizer: true,
  api_id: options.apiId,
  api_hash: options.apiHash,
  database_directory: options.databaseDirectory,
  files_directory: options.filesDirectory,
  use_test_dc: options.useTestDc
}

In a real application, you probably want to change device_model and other parameters.

client.login(fn?: () => LoginDetails) => Promise<void>

Log in to your Telegram account.

await client.login()

By default, tdl asks the user for the phone number, auth code, and 2FA password (if needed) in the console. You can override the defaults with custom functions, for example:

// Example
await client.login(() => ({
  getPhoneNumber: retry => retry
    ? Promise.reject('Invalid phone number')
    : Promise.resolve('+9996620001'),
  getAuthCode: retry => retry
    ? Promise.reject('Invalid auth code')
    : Promise.resolve('22222'),
  getPassword: (passwordHint, retry) => retry
    ? Promise.reject('Invalid password')
    : Promise.resolve('abcdef'),
  getName: () => Promise.resolve({ firstName: 'John', lastName: 'Doe' })
}))

The getName function is called if the user is not signed up.

client.login supports only a subset of authentication methods available on Telegram. It is possible (and advisable for larger apps) not to use the client.login helper and implement the authorization process manually, handling authorizationStateWaitPhoneNumber and other updates.

This function accepts the following interface:

type LoginDetails = {
  type?: 'user',
  getPhoneNumber?: (retry?: boolean) => Promise<string>,
  getEmailAddress?: () => Promise<string>,
  getEmailCode?: () => Promise<string>,
  confirmOnAnotherDevice?: (link: string) => void,
  getAuthCode?: (retry?: boolean) => Promise<string>,
  getPassword?: (passwordHint: string, retry?: boolean) => Promise<string>,
  getName?: () => Promise<{ firstName: string, lastName?: string }>
} | {
  type: 'bot',
  getToken: (retry?: boolean) => Promise<string>
}
// Note that client.login accepts a function that returns the object, not the
// object directly. The function will not be called if the client is already
// authorized.
declare function login (fn?: () => LoginDetails): Promise<void>

getEmailAddress and getEmailCode are called in TDLib >= v1.8.6 only.

client.loginAsBot(token: string | (() => string | Promise<string>)) => Promise<void>

Instead of logging in as a user, you can log in as a bot by token.

await client.loginAsBot('YOUR_BOT_TOKEN') // Enter your token from @BotFather

client.on(event: string, callback: Function) => Client

Attach an event listener to receive updates and other events.

function onUpdate (update) {
  console.log('New update:', update)
}
client.on('update', onUpdate)
client.on('error', console.error)

Ideally, you should always have a listener on client.on('error'). There is no default listener, all errors will be ignored otherwise.

You can consider using reactive libraries like RxJS or most.js for convenient event processing.

Some other rarely-used events also exist and are described in the TypeScript interface.

client.addListener is an alias for client.on.

client.once(event: string, callback: Function) => Client

Attach a one-time listener.

client.off(event: string, listener: Function, once?: boolean) => Client

Remove an event listener.

const listener = u => {
  console.log('New update:', u)
  if (u?.authorization_state?._ === 'authorizationStateReady')
    client.off('update', listener) // Removes the listener
}
client.on('update', listener)

client.removeListener is an alias for client.off.

client.invoke(query: Object) => Promise<Object>

Call a TDLib method asynchronously. The promise can be rejected with a TDLib object of type _: 'error'.

For the information regarding TDLib API list, see the "Getting started" section of this README.

const chats = await client.invoke({
  _: 'getChats',
  chat_list: { _: 'chatListMain' },
  limit: 4000
})

await client.invoke({
  _: 'sendMessage',
  chat_id: 123456789,
  input_message_content: {
    _: 'inputMessageText',
    text: {
      _: 'formattedText',
      text: '👻'
    }
  }
})

client.close() => Promise<void>

Close the TDLib client.

await client.close()

tdl.execute(query: Object) => (Object | null)

Call a TDLib method synchronously. This function can be used only with the methods marked as "can be called synchronously" in the TDLib documentation.

const res = tdl.execute({
  _: 'getTextEntities',
  text: '@telegram /test_command https://telegram.org telegram.me'
})

client.execute is the same as tdl.execute.

tdl.setLogMessageCallback(maxVerbosityLevel: number, cb: Function | null) => void

Set the callback that is called when a message is added to the TDLib log. This corresponds to the td_set_log_message_callback tdjson function.

Types

While tdl works with any TDLib version (above the requirement), the TypeScript types have to be installed specifically for the TDLib version you use. This can be done via a small tdl-install-types utility, which downloads and generates types for you. It can be called using npx tdl-install-types without manually installing.

$ npx tdl-install-types [<options>] [<target>]

It can generate types given a tdjson library (e.g. npx tdl-install-types ./libtdjson.so), a TDLib git ref (examples: npx tdl-install-types v1.8.0, npx tdl-install-types master, npx tdl-install-types 2de39ffffe71dc41c538e66085658d21cecbae08), or a td_api.tl file (npx tdl-install-types td_api.tl). When called without arguments, it will try to use require('prebuilt-tdlib').getTdjson() as the target. By default, the types are generated into a tdlib-types.d.ts file that you can git-commit.

See npx tdl-install-types@latest --help for additional information.

The types can be imported by using the tdlib-types module name:

import type * as Td from 'tdlib-types'
// And use as: Td.message, Td.user, ...

It is considerably more convenient to use tdl with TypeScript, which enables full autocompletion for the TDLib methods and objects along with the documentation.

Note that when using npx, the version of tdl-install-types might be outdated if you are not appending the @latest tag. You can also install the utility globally or per-project as a dev dependency.

Other JavaScript runtimes

Since bun is Node.js-compatible and supports Node-API, tdl should work out of the box, however the stability may not be the best yet.

deno can also import tdl through the node compatibility via import * as tdl from 'npm:tdl'. Currently, the Node-API implementation in deno seems to be unstable and random segfaults sometimes occur. Not recommended to use tdl in deno.

Possible errors

• UPDATE_APP_TO_LOGIN

Update TDLib to v1.7.9 (v1.8.0) or newer. It is no longer possible to log in by phone number in older versions of TDLib.

• Dynamic Loading Error: Win32 error 126 (Windows)
• Dynamic Loading Error: dlopen(…) image not found (macOS)
• …cannot open shared object file: No such file or directory (Linux)

The tdjson shared library or one of its dependencies (for example, libssl) cannot be found. To troubleshoot dependency issues, try to run ldd libtdjson.so on Linux or otool -L libtdjson.dylib on macOS. On Windows, you can use an app like Dependency Walker.

Recheck the documentation of dlopen (Linux), dlopen (macOS), Dynamic-Link Library Search Order (Windows) to make sure the shared library is present in the search paths. By default, Linux does not search in the current working directory, while macOS does.

• fatal error: napi.h: no such file or directory
• error: no such file or directory: …/node-modules/node-addon-api

The path to the directory where you execute npm install likely contains spaces, which is not supported by gyp: https://github.com/nodejs/node-gyp/issues/65#issuecomment-368820565.

• Error while reading RSA public key

You can get this error if libtdjson is dynamically linked against OpenSSL and some of the symbols got resolved into Node.js instead of the system OpenSSL.

Note that Node.js also uses OpenSSL (the distributed binaries are statically linked against it) and exports the OpenSSL symbols. In the result, there are two versions of OpenSSL in the same application. Then, using standard dlopen, especially on Linux, most of the symbols will be resolved into libcrypto inside the Node.js binary, not into the system libcrypto. It still can work correctly if the versions are ABI-compatible, i.e. if TDLib is linked against an OpenSSL version sufficiently similar to the version that Node.js uses (node -p "process.versions.openssl").

tdl tries to get around the symbol conflict issues by using RTLD_DEEPBIND when available, so these issues should be rare in practice.

You can use lldb or gdb to check whether the symbols get resolved into Node.js. For example, open lldb -- node index.js and set these breakpoints:

break set -r EVP_ -s node
break set -r AES_ -s node
break set -r BIO_ -s node
break set -r RSA_ -s node
break set -r CRYPTO_ -s node

It's also possible to set breakpoints inside the system OpenSSL:

break set -r . -s libcrypto.so.1.1
break set -r . -s libssl.so.1.1

To solve this issue, try to link TDLib statically against OpenSSL (the OPENSSL_USE_STATIC_LIBS option in cmake) or link it against the OpenSSL version that Node.js uses.

Another possible option is to rebuild Node.js from source, linking it dynamically against the same system OpenSSL. That way, there is only one instance of OpenSSL in the application. For example, using nvm, you can install Node.js v16 from source on GNU/Linux via this command:

$ nvm install -s 16 --shared-openssl --shared-openssl-includes=/usr/include/ --shared-openssl-libpath=/usr/lib/x86_64-linux-gnu/

However, it's inconvenient for most users to rebuild Node.js.

Another hypothetical solution is to rebuild TDLib with the OpenSSL headers distributed in Node.js (<path-to-node>/include/node/) without linking it to anything, simply leaving the undefined symbols. Using this option, there is also only one OpenSSL. I haven't checked that this works or that Node exports all the symbols needed for TDLib. With this option, TDLib also should be rebuilt every time Node.js updates the OpenSSL dependency.

This issue doesn't apply to Electron because it doesn't export the OpenSSL symbols.

• Segmentation fault

The cause of the segfault might be the same as above.